OBJECTIVE To determine using a routine clinical program the time course of the fatigue effect in both eyes at a given examination, the influence of rest periods during the examination, and the differences in the fatigue effect between persons with ocular hypertension and age-matched normal subjects. METHODS Perimetry was undertaken for both eyes of 20 normal persons and 20 persons with ocular hypertension using Program G1X (Interzeag AG, Schlieren, Switzerland) of the Octopus 1-2-3 perimeter. Global mean defect and loss variance were calculated for each of the eight individual stages and global SF was calculated for stages 5 to 8. The superoinferior and nasotemporal hemifield indices and those for within and beyond 17 degrees eccentricity were calculated separately to investigate the locus of the fatigue effect. RESULTS Group mean mean defect and loss variance deteriorated over stages (P < 0.001) and between eyes for both groups (P < 0.001; P < 0.004). The deterioration in the mean defect was more marked peripherally (P < 0.001) and inferiorly with age and group (P < 0.001) whereas the localized loss was more pronounced peripherally (P < 0.001), superiorly (P < 0.010) and nasally (P < 0.001). A 1-minute break during the examination of a given eye and a 3-minute break between eyes was insufficient to overcome the fatigue effect. The performance of the two groups was almost indistinguishable. CONCLUSIONS The deterioration in the visual field indices mean defect and loss variance challenges the currently accepted ideal length of a perimetric examination with the conventional algorithms. Confidence limits for the definition of abnormality with these algorithms should reflect the presence of the fatigue effect and be different for the second eye.